The Impact of Cash Transfers on the Educational Attainment, Sexual Behavior, and HIV Status of Adolescent Girls in Malawi

Sarah Baird
Berk Özler
Ephraim Chirwa
Richard Garfein
Fieldwork by:
Zomba District, Malawi
2,907 schoolgirls
2007 - 2007
Target group:
  • Secondary schools
  • Students
  • Women and girls
Outcome of interest:
  • Communicable diseases
  • Enrollment and attendance
  • Sexual and reproductive health
Intervention type:
  • Conditional cash transfers
AEA RCT registration number:

Cash transfers are popular development programs, but little is known about their long-term impacts and whether conditionality is essential to achieving positive outcomes in health and education. In Malawi, researchers tested the relative effects of providing conditional and unconditional cash transfers to teenage girls and their families. Among girls who were in school at the start of the program, conditional cash transfers increased school attendance and reduced HIV prevalence while unconditional transfers were more effective in helping girls delay marriage and childbearing.

Policy issue

Conditional cash transfers (CCTs) are popular tools for development: over 29 countries have started CCT programs since the 1990s. By conditioning payments on activities that are widely seen as beneficial, such as attending school or regular health checkups, these programs aim to increase these activities and generate improvements in associated long-term outcomes. For instance, schooling has been shown to increase learning, improve labor market outcomes, and cause girls to delay childbearing, so a program that increases attendance should also impact outcomes such as test scores, wages, and pregnancy rates. A large body of evidence has shown that CCTs are effective in improving the direct, short-term outcomes on which they are conditioned, but it is less clear whether CCT programs have a positive impact on indirect outcomes. It is also unclear how important conditionality is to changing recipients’ behavior. It is possible that unconditional cash transfers (UCTs) can generate improvements in health and education without the high costs of monitoring recipients’ compliance with various conditions.

This study in Malawi compares conditional and unconditional cash transfers and looks at both direct effects such as education and indirect outcomes such as HIV prevalence, making it one of the first to untangle the impact of conditionality in cash transfer programs.

Context of the evaluation

Malawi is among the poorest countries in Sub-Saharan Africa. More than three-quarters of its population live in rural areas and most are subsistence farmers. The per capita income in 2008 was US$760, far below an average of US$1,973 in the region. The country also performs poorly on measures of education and health. Only 24 percent of secondary school-aged youth were enrolled in school in 2008. Fourteen percent of the adult population was infected with HIV, and among young adults, prevalence was more than four times higher for women than for men.

This study took place in Zomba district in southern Malawi, which has high levels of poverty, low school enrollment, and high HIV prevalence, even in comparison to the rest of the country. The district has a large urban center, Zomba city, and many rural villages.

Twenty Malawi women pose for picture
Young women in Zomba district of Malawi
Sarah Baird

Details of the intervention

Researchers tested the effect of a CCT program on a range of direct and indirect outcomes, including school enrollment, attendance, marriage, childbearing, sexual behavior, and prevalence of sexually transmitted diseases. They also compared conditional and unconditional transfers to test the effect of conditioning the payment on school attendance.

Researchers selected 176 enumeration areas (EAs), or administrative subdivisions containing around 250 households, from urban, semi-rural, and rural areas of Zomba district for the study. In each EA, a portion of unmarried girls aged 13 to 22 was randomly selected to participate. Half of the EAs were assigned to receive cash transfers while the other half served as a comparison group that did not receive transfers.

The 88 EAs that were assigned to receive cash transfers were then further divided to receive different types of transfers. Girls who were in school at the start of the study (“baseline schoolgirls”) were offered conditional transfers in 46 EAs, unconditional transfers in 27 EAs, and no transfers in 15 EAs. In all 88 of the treatment EAs, girls who had dropped out of school at the start of the study (“baseline dropouts”) were offered cash transfers conditional on attending school.

In total, 3,796 girls (2,907 baseline schoolgirls and 889 baseline dropouts) were interviewed for the study. Girls who received a transfer were randomly assigned an amount of US$1, $2, $3, $4, or $5 per month. The girls’ parents received a transfer of US$4, $6, $8, or $10 per month, which randomly varied by EA. The program also paid secondary school fees for girls in the CCT group. To match this, the transfer amounts for secondary school girls in the UCT group were adjusted upward by the average secondary school fee, but were otherwise identical.

In the CCT group, girls had to attend 80 percent of the days school was in session each month to receive that month’s transfer. Attendance was never checked for girls in the UCT group, and they received their payments by showing up at a transfer location each month. Even though girls in the UCT group generally understood that there were no requirements for them to receive the transfer, they knew that the program was intended to promote education. Both transfer programs ran for the ten months of the school year over two years, or from January 2008 to December 2009.

Results and policy lessons

Researchers are still analyzing data for a complete analysis—the results reported below focus on outcomes for baseline schoolgirls only.

Researchers found that CCTs to teenage girls and their families improved a wide range of outcomes, including school attendance, sexual behavior, and disease prevalence. The unconditional variation of the program, which did not impose attendance requirements for receiving payment, was more effective in helping baseline schoolgirls delay marriage and childbearing. This suggests that policymakers may have to consider the trade-off between education and marriage/fertility decisions when designing cash transfer programs.

School enrollment, attendance, and test scores: The CCT program improved schooling outcomes more effectively than the UCT program. Among girls who were enrolled in school at baseline, those in the CCT group were enrolled for 0.54 more terms throughout the six-term program, relative to a comparison group average of 4.8 terms. Girls in the UCT group were enrolled for 0.23 additional terms. Attendance was 8 percentage points higher in the CCT group, relative to a comparison group average of 81 percent. The UCT program did not significantly affect attendance. Girls in the CCT group performed better on tests of English reading comprehension and cognitive ability, but the UCT program had no effect on test scores.

Marriage and childbearing: The UCT program was more effective than the CCT program in delaying marriage and childbearing among baseline schoolgirls. Among these girls, 18 percent of the comparison group had married after two years. Girls in the CCT group were just as likely to be married, but girls in the UCT group were nearly 8 percentage points less likely to be married. About a quarter (24.7 percent) of the comparison group became pregnant during the program. Girls in the CCT group were equally likely to become pregnant, but girls in the UCT group were nearly 7 percentage points less likely to become pregnant. It appears that the unconditional transfer allowed girls who dropped out of school–and therefore would have stopped receiving conditional payments–to support themselves without relying on a husband or having transactional sex with older men, thereby delaying marriage and childbearing.

Sexual behavior: Both cash transfer programs reduced sexual behavior among baseline schoolgirls. Only 3 percent of girls receiving CCTs or UCTs reported having sex at least once per week, compared to 7 percent of girls in the comparison group. Girls receiving transfers were also significantly less likely to have had a partner over age 25 (less than 1 percent of girls receiving transfers compared to 3 percent of the comparison group).

Sexually transmitted diseases: Researchers tested a subset of girls for HIV and herpes simplex virus type 2 (HSV-2), a sexually transmitted infection more common than HIV, after 18 months and found that the transfer program decreased HIV and HSV-2 prevalence among baseline schoolgirls. Among these girls, 1.2 percent of those who were offered either CCTs or UCTs tested positive for HIV, compared with 3 percent of girls in the comparison group. The HSV-2 prevalence was less than 1 percent among girls offered either CCTs or UCTs and 3 percent in the control group.

In J-PAL's comparative cost-effectiveness analyses, the UCT program had no impact on test scores but led to 0.02 additional years of education per $100 spent. The CCT program led to a 0.06 standard deviation improvement in test scores and 0.09 additional years of education per $100 spent for those who received the minimum transfer amount, and 0.07 additional years of education per $100 spent for those who received the average transfer amount. For more information, see the full comparative cost-effectiveness analyses for test score performance and student participation.

Baird, Sarah, Craig McIntosh, and Berk Özler. 2011. “Cash or Condition? Evidence from a Randomized Cash Transfer Program.” Quarterly Journal of Economics 126 (4): 1709-1753.

Baird, Sarah, Ephraim Chirwa, Craig McIntosh, and Berk Özler. 2010. "The Short-Term Impacts of a Schooling Conditional Cash Transfer Program on the Sexual Behavior of Young Women." Health Economics 19(2010): 55-68.

Baird, Sarah J, Richard S Garfein, Craig T McIntosh, and Berk Özler. 2012. "Effect of a Cash Transfer Programme for Schooling on Prevalence of HIV and Herpes Simplex Type 2 in Malawi: A Cluster Randomised Trial." Lancet 379(2012): 1320-1329.